April 2026 - talks - lists.cs.princeton.edu

Fangqi Dong will present "Explicit Separations for One-Query Unitary Synthesis" on 4/21/26 at 1:30 PM in CS 301.
by Nicki Mahler 21 Apr '26

21 Apr '26

Fangqi Dong will present "Explicit Separations for One-Query Unitary Synthesis" on 4/21/26 at 1:30 PM in CS 301. Committee: Alex Lombardi(adviser), Ran Raz, Pravesh Kothari Abstract & Reading List: [ https://docs.google.com/document/d/1LiUZ2vXIhlSFCytJqTmMB5XQkjcHlBQ9Xq-m0rA… | https://docs.google.com/document/d/1LiUZ2vXIhlSFCytJqTmMB5XQkjcHlBQ9Xq-m0rA… ] Abstract: The unitary synthesis problem (Aaronson-Kuperberg CCC’07) asks whether an arbitrary n-qubit unitary U can be implemented by a polynomial size quantum circuit, given access to some classical oracle f_U that depends on U. Recently, it was proved (Lombardi-Ma-Wright STOC’24) that for a Haar-random unitary, this is impossible for any efficient synthesis algorithm that makes only one query (or polynomially many parallel queries) to an arbitrary classical oracle. In this work, we extend this line of lower bound for random unitaries to a finer structural understanding of one-query synthesis for natural explicit families. In particular, we prove one-query lower bounds for explicit families of unitaries that have efficient two-query synthesis algorithms, thereby separating the power of one-query and two-query synthesis algorithms. We also give a one-query algorithm for complex phase unitaries that achieves constant approximation, and prove a sharper separation between one-query unitary synthesis and quantum programs.

2 3

Frederick Qiu will present his General Exam "Separations between Algorithms and Truthful Mechanisms" on April 27, 2026 at 12pm in CS 302.
by Gradinfo 20 Apr '26

20 Apr '26

Frederick Qiu will present his General Exam " Separations between Algorithms and Truthful Mechanisms" on April 27, 2026 at 12pm in CS 302. Zoom Link: [ https://princeton.zoom.us/j/3603536257 | https://princeton.zoom.us/j/3603536257 ] Committee: Matthew Weinberg(adviser), Mark Braverman, and Huacheng Yu Link to Abstract and Reading List: [ https://docs.google.com/document/d/1FtdsKcw3zmFtl7DfMWENuEFNuUfa-FFOgf-ynmn… | https://docs.google.com/document/d/1FtdsKcw3zmFtl7DfMWENuEFNuUfa-FFOgf-ynmn… ] Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

2 2

Kylie Zhang will present her MSE talk "AI-Assisted Moot Courts: Simulating Justice-Specific Questioning in Oral Arguments" on Weds 4/22 at 10am in Sherrerd 306.
by Gradinfo 20 Apr '26

20 Apr '26

Kylie Zhang will present her MSE talk "AI-Assisted Moot Courts: Simulating Justice-Specific Questioning in Oral Arguments" on Weds 4/22 at 10am in Sherrerd 306. Committee: Prof. Peter Henderson (adviser) and thesis reader is Prof. Jonathan Mayer. Abstract: "In oral arguments, judges probe attorneys with questions about the factual record, legal claims, and the strength of their arguments. To prepare for this questioning, both law schools and practicing attorneys rely on moot courts: practice simulations of appellate hearings. Leveraging a dataset of U.S. Supreme Court oral argument transcripts, we examine whether AI models can effectively simulate justice-specific questioning for moot court-style training. Evaluating oral argument simulation is challenging because there is no single correct question for any given turn. Instead, effective questioning should reflect a combination of desirable qualities, such as anticipating substantive legal issues, detecting logical weaknesses, and maintaining an appropriately adversarial tone. We introduce a two-layer evaluation framework that assesses both the realism and pedagogical usefulness of simulated questions using complementary proxy metrics. We construct and evaluate both prompt-based and agentic oral argument simulators. We find that simulated questions are often perceived as realistic by human annotators and achieve high recall of ground truth substantive legal issues. However, mod- els still face substantial shortcomings, including low diversity in question types and sycophancy. Importantly, these shortcomings would remain undetected under naive evaluation approaches."

1 0

Daniel Williams will present his General Exam "Daniel Williams will present his General Exam "An Efficient and Scalable Language Design for High-Dimensional Clifford Algebras on GPU" on Wednesday, April 22, 2026 at 3:00 PM in E-Quad J223 and via zoom.
by CS Grad Department 17 Apr '26

17 Apr '26

Daniel Williams will present his General Exam "Daniel Williams will present his General Exam "An Efficient and Scalable Language Design for High-Dimensional Clifford Algebras on GPU" on Wednesday, April 22, 2026 at 3:00 PM in E-Quad J223 and via zoom. Zoom link: https://princeton.zoom.us/j/6762826001 Committee Members: Ryan Adams (advisor), David Walker, Christine Allen-Blanchette Abstract: Many of the most challenging problems in computational physics and engineering, ranging from electromagnetism to inverse design of complex mechanical systems, are fundamentally geometric. However, conventional linear algebra based numerical methods do not natively encode geometric structure, often leading to ill-conditioned optimization, constraint drift, and poor scaling in high-dimensional settings. While Clifford Algebras provide a principled and coordinate-free mathematical framework for representing symmetries, high dimensional spaces, and differential geometry on curved manifolds, its practical adoption in large-scale scientific computing has been limited by exponential memory requirements and lack of GPU-accelerated infrastructure. This work introduces a new framework for accelerated computational geometry in high dimensions using Clifford Algebras. We develop a static type system that automatically infers and reduces memory constraints, allowing numerical execution to operate on dramatically reduced representations without approximation. Building on this foundation, we implement a Python-based Just-in-Time (JIT) compiler integrated with JAX, the accelerated array computation library, that traces user-defined geometry programs of various clifford algebra signatures and lowers them into optimized GPU kernels, while preserving precision and full compatibility with automatic differentiation. Preliminary results demonstrate substantial improvements in tractable dimensionality and GPU throughput under batching, extending feasible computation from n < 6 to up to n = 20 dimensions on a single GPU by exploiting known sparsity before execution. Reading List: https://docs.google.com/document/d/1eSBhk0Irmfqf26QOlY5Jrbb_kUmhOQgN4Jbay6_… Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so.

1 0

Nobline Yoo will present her MSE Talk "D2D: Detector-to-Differentiable Critic for Improved Numeracy in Text-to-Image Generation" on Monday, April 20, 2026 at 12:30p in CS 402.
by CS Grad Department 17 Apr '26

17 Apr '26

Nobline Yoo will present her MSE Talk "D2D: Detector-to-Differentiable Critic for Improved Numeracy in Text-to-Image Generation" on Monday, April 20, 2026 at 12:30p in CS 402. Committee Members: Prof. Olga Russakovsky (advisor), Prof. Zhuang Liu (reader). All are welcome to attend. Title: D2D: Detector-to-Differentiable Critic for Improved Numeracy in Text-to-Image Generation Abstract: Text-to-image (T2I) diffusion models have achieved strong performance in semantic alignment, yet they still struggle with generating the correct number of objects specified in prompts. Existing approaches typically incorporate auxiliary counting networks as external critics to enhance numeracy. However, since these critics must provide gradient guidance during generation, they are restricted to regression-based models that are inherently differentiable , thus excluding detector-based models with superior counting ability, whose count-via-enumeration nature is non-differentiable . To overcome this, we propose Detector-to-Differentiable ( D2D ), a novel framework that transforms non-differentiable detection models into differentiable critics, leveraging their superior counting ability to guide numeracy generation. Specifically, we design custom activation functions to convert detector logits into soft binary indicators, which are then used to optimize the noise prior at inference time with pre-trained T2I models. Our extensive experiments on SDXL-Turbo, SD-Turbo, and Pixart-DMD across four benchmarks of varying complexity (low-density, high-density, and multi-object scenarios) demonstrate consistent and substantial improvements in object counting accuracy (e.g., boosting up to 13.7% on D2D-Small, a 400-prompt, low-density benchmark), with minimal degradation in overall image quality and computational overhead.

1 0

Max Gonzalez Saez-Diez will present his MSE thesis "D3: Dataset, Diagnosis, and Deployment of Video Models for Police Body-Worn Camera Understanding" on Monday, April 20th, 2026 in CS 402 at 12:00pm.
by CS Grad Department 17 Apr '26

17 Apr '26

Max Gonzalez Saez-Diez will present his MSE thesis "D3: Dataset, Diagnosis, and Deployment of Video Models for Police Body-Worn Camera Understanding" on Monday, April 20th, 2026 in CS 402 at 12:00pm. Advisor: Peter Henderson Reader: Olga Russakovsky All are welcome to attend. Abstract: Vision-language models are increasingly applied to police body-worn camera (BWC) video, yet few public benchmarks exist and researchers have little means to evaluate private vendor claims. This thesis addresses all three gaps: constructing the data necessary for training, evaluating models in a zero-shot manner on long-tailed phenomena, and demonstrating how this data can be used to train and deploy models in practice. First, we introduce EgoPolice. To the best of our knowledge, it is the largest existing dataset for egocentric understanding of real body-worn camera footage. Second, we present BWC-Bench, a zero-shot captioning benchmark that scores whether each listed action is asserted in the generated caption. Third, we describe work with a partner agency that provided access to their videos and use-of-force reports. To the best of our knowledge, this constitutes the first analysis of real police body-worn camera footage and associated use-of-force reports at this scale. A physical-interaction detector trained on EgoPolice ranks clips from a large private archive for human review. We find that models can support human-in-the-loop analysis at a scale not previously possible and uncover evidence of substantial underreporting of use-of-force events.

1 0

Wuwei Zhang will present her MSE Talk"Improving Long-Context Reasoning with Query-Focused Retrieval Heads" onTuesday, April 21, 2026 at 4:00p in CS 302.
by CS Grad Department 16 Apr '26

16 Apr '26

Wuwei Zhang will present her MSE Talk "Improving Long-Context Reasoning with Query-Focused Retrieval Heads" on Tuesday, April 21, 2026 at 4:00p in CS 302. Committee Members: Prof. Danqi Chen (advisor),Prof. Karthik Narasimhan (reader). All are welcome to attend. Title: Improving Long-Context Reasoning with Query-Focused Retrieval Heads Abstract: Large language models (LLMs) have demonstrated strong performance across many tasks, yet they often struggle to effectively utilize information from long contexts. Recent studies suggest that a subset of attention heads, known as retrieval heads, plays a critical role in retrieving relevant information from long contexts. However, existing methods for identifying these heads rely on copy-paste behavior in synthetic tasks, which may not reflect real-world retrieval requirements. In this thesis, we investigate how retrieval heads contribute to long-context reasoning and propose methods to better utilize them during inference. First, we introduce Query-Focused Retrieval Heads (QRHead), a method for identifying attention heads that selectively retrieve information relevant to the input query. By aggregating the accumulated attention mass of QRHead, we further develop QRRetriever, a training-free retrieval approach that selects the most relevant context segments for downstream reasoning tasks. Building on QRHead, we propose Dynamic Attention-Scaling Decoding (DySCO), an inference-time method that dynamically rescales attention based on query-focused retrieval signals, enabling models to more effectively access relevant information during generation. Our methods achieve substantial improvements on long-context reasoning benchmarks across multiple model families, without requiring additional training. Together, these results demonstrate that retrieval heads provide a useful lens for understanding and improving long-context reasoning in LLMs, and that inference-time interventions guided by these mechanisms can lead to more effective context utilization.

1 0

Amelia Dobis will present her General Exam "New Paradigms for Hardware Design, Compilation, and Verification" on April 24, 2026 at 1pm at 194 Nassau St, 2nd floor meeting area (Suite 21) and Zoom.
by Gradinfo 15 Apr '26

15 Apr '26

Amelia Dobis will present her General Exam "New Paradigms for Hardware Design, Compilation, and Verification" on April 24, 2026 at 1pm at 194 Nassau St, 2nd floor meeting area (Suite 21) and Zoom. Committee: Mae Milano (adviser), Aarti Gupta, Sneha Goenka Zoom: [ https://princeton.zoom.us/j/99419002211?pwd=hd5f2sSwZUt7yfU3bBTNOILyAKz6tE.1 | https://princeton.zoom.us/j/99419002211?pwd=hd5f2sSwZUt7yfU3bBTNOILyAKz6tE.1 ] Everyone is invited to attend the talk, and those faculty wishing to remain for the oral exam following are welcome to do so. [ https://docs.google.com/document/d/1TdYi6TAy0gCCj46ryReP9Ms9C-ila6dQ8dNhuCF… | Abstract & Reading List: https://docs.google.com/document/d/1TdYi6TAy0gCCj46ryReP9Ms9C-ila6dQ8dNhuCF… ]

1 0

Sofiia Druchyna will present her MSE Talk "Spectral Credit Reconstruction for GRPO in Long-Horizon RLHF" on Monday, April 20, 2026 at 10:30a in CS 401.
by CS Grad Department 15 Apr '26

15 Apr '26

Sofiia Druchyna will present her MSE Talk "Spectral Credit Reconstruction for GRPO in Long-Horizon RLHF" on Monday, April 20, 2026 at 10:30a in CS 401. Committee Members: Prof. Elad Hazan (advisor), Prof. Manoel Horta Ribeiro (reader). All are welcome to attend. Title: Spectral Credit Reconstruction for GRPO in Long-Horizon RLHF Abstract: Reinforcement Learning from Human Feedback (RLHF) relies on sequence-level rewards to train large language models, yet policy updates are typically applied uniformly across all tokens in a generated completion. This uniform credit assignment ignores the temporal structure of language generation and contributes to high-variance gradient estimates in long-horizon settings. We propose a spectral credit assignment method for Group-Relative Policy Optimization (GRPO) that redistributes scalar advantages across token positions using a low-dimensional spectral parameterization. Our approach represents token-level credit as a smooth combination of fixed low-frequency basis functions, with mixture weights predicted by a state-dependent gating network operating on pooled hidden states. This construction preserves the expected GRPO update while constraining token-level variation to a controlled, low-frequency subspace, thereby reducing variance without introducing additional critics or token-level rewards. We show that our method interpolates smoothly between standard GRPO and structured token-level credit assignment, offering a simple and stable mechanism for incorporating temporal structure into RLHF training.

1 0

Phyllis Wang will present her MSE talk "Implementation and Scaling of Quantum Cognitive Models" on Thursday, April 23rd, 2026 at 1pm in CS 301.
by Gradinfo 15 Apr '26

15 Apr '26

Phyllis Wang will present her MSE talk "Implementation and Scaling of Quantum Cognitive Models" on Thursday, April 23rd, 2026 at 1pm in CS 301. Adviser: Margaret Martonosi, Reader: Raghav Pothukuchi All are welcome to attend Title: Implementation and Scaling of Quantum Cognitive Models Abstract: The computational aspect of cognitive models inspired by quantum mechanics poses unique considerations in their implementation, both through classical and quantum computing. We examine the quantum well-inspired Multi-Particle Multi-Well (MPMW) model for decision making to find how quantum cognitive techniques scale on various types of hardware and modeling approaches. This thesis explores an end-to-end implementation of the MPMW model, including the representation of quantum wells as the landscape becomes more discretized, addressing the challenge of memory bottlenecks that arise from representing quantum states. We further explore the computational and memory scaling limits of finding the eigenstates of the wells, and analyze the results of scaling on the model predictions. Finally, we look at the role of quantum computing and the possibility of scaling model sizes further through the use of the hybrid quantum-classical Variational Quantum Eigensolver algorithm.

1 0